Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!yale!cmcl2!lanl!cwi
From: cwi@lanl.gov (Craig W Idler)
Newsgroups: comp.benchmarks
Subject: SPEC benchmarks
Keywords: how its done.
Message-ID: <5970@lanl.gov>
Date: 15 Nov 90 00:14:56 GMT
Organization: Los Alamos Natl Lab, Los Alamos, N.M.
Lines: 10

Taken from and article in EE times...

The SPEC benchmarks consist of ten tests, mainly large application
programs. Of these, six are concerned with floating-point performance; the
remaining four with integer performance. The time a system takes to run
each test is compared with the time it takes a standard DEC VAX 11/780
(1 Mips machine) to do it.  This yields a SPECratio for each test.
  These separate ratios can be used to generate a single performance
number - the SPECmark. A SPECmark is the geometric mean (the 10th root)
of the product of the 10 SPECratios.